Socioeconomic inequalities in life and health

Ageing

RESEARCH REPORT

Socioeconomic inequalities in life and health expectancies around official retirement age in 10 Western-European countries I M Majer, W J Nusselder, J P Mackenbach, A E Kunst Department of Public Health, Erasmus MCdUniversity Medical Center Rotterdam, Rotterdam, The Netherlands Correspondence to I M Majer, Department of Public Health, Erasmus MC e University Medical Center Rotterdam, PO Box 2040, 3000 CA Rotterdam, the Netherlands; [email protected] Accepted 21 October 2010 Published Online First 23 November 2010

ABSTRACT Background Discussions on raising pension eligibility age focus more on improvement in life expectancy (LE) and health expectancy measures than on socioeconomic differences in these measures. Therefore, this study assesses the level of socioeconomic differences in these two measures in Western-Europe. Methods Data from seven annual waves (1995e2001) of the European Community Household Panel were used. Health and socioeconomic information was collected using standardised questionnaires. Health was measured in terms of disability in daily activities. Socioeconomic status was determined as education level at baseline. Multi-state Markov modelling was applied to obtain agespecific transition rates between health states for every country, educational level and gender. The multi-state life table method was used to estimate LE and disability free life expectancy (DFLE) according to country, educational level and gender. Results When comparing high and low educational levels, differences in partial DFLE between the ages 50 and 65 years were 2.1 years for men and 1.9 years for women. At age 65 years, for LE the difference between high and low educated groups was 3 years for men and 1.9 years for women, and for DFLE the difference between high and low educated groups was 4.6 years for men and 4.4 years for women. Similar patterns were observed in all countries, although inequalities tended to be greater in the southern countries. Conclusions Educational inequalities, favouring the higher educated, exist on both sides of the retirement eligibility age. Higher educated persons live longer in good health before retirement and can expect to live longer afterwards.

INTRODUCTION Social policymakers in Western-European countries are facing a common problem with regard to population ageing. Low birth rates, increasing life expectancy (LE) and dependency ratios have resulted in increased spending on pensions. Many countries have undertaken systematic restructuring of their pension system, including adjustment of the pension eligibility age to increasing LE.1 2 At present, in most Organization for Economic Cooperation and Development (OECD) countries, the eligibility age for retirement among men and women is 65 years. Although the rationale for such restructuring is to improve the financial sustainability of pension systems, such reforms may have adverse social 972

effects.1 2 An increase in LE is not necessarily equivalent to being able to work longer and the extent to which total LE increases may differ from the increase in health expectancy (HE).3 4 Furthermore, both LE and HE are strongly related to socioeconomic status (SES). If the standard retirement age is raised, people at the bottom of the socioeconomic ladder could be disproportionately affected. Whether one should consider socioeconomic differences in LE and HE in discussions on raising the pension eligibility age depends on the magnitude of these differences. Therefore, it is important to establish to what extent people in various socioeconomic groups are healthy and remain healthy at older ages. Estimates of socioeconomic differences in HE are available for an increasing number of European countries.5e11 Unfortunately, estimates from national studies could not be compared due to large variations in the data sources used, the age ranges covered, and the health and socioeconomic indicators used. Because no study has used data from more than two countries to report on LE and HE by SES, an overview of the magnitude of socioeconomic differences in total and healthy life years at old age in Europe is still lacking. Therefore, this study aims to determine the magnitude of socioeconomic differences in LE and HE, measured in terms of disability free life expectancy (DFLE) in European countries: more specifically, inequalities in DFLE between the ages of 50 and 65 years, and in LE and DFLE after the age of 65 years. DFLE combines information on mortality and disability into a summary measure of the expected number of years to be lived without disability. Investigation of these measures at these specific age intervals provides data on several aspects relevant to discussions on pension age.

METHODS Data The data for this study were derived from the European Community Household Panel (ECHP). The ECHP is a standardised multi-purpose annual longitudinal social survey carried out at the level of the European Union (EU) between 1994 and 2001. It is centrally designed and coordinated by the Statistical Office of the European Communities (Eurostat) and covers demographics, labour force behaviour, income, health, education and training, housing, migration, etc. The data were collected by the National Statistical Institutes or research centres of the participating countries using

J Epidemiol Community Health 2011;65:972e979. doi:10.1136/jech.2010.111492

Ageing a uniform random sampling design and common blueprint questionnaires. Data checks, imputation and weighting were done centrally to maximise data quality. ECHP aimed at being both cross-sectionally and longitudinally representative for the national populations. The design and procedures of the ECHP have been extensively reviewed in Peracchi.12 Information on non-response rates at baseline is given by Huisman et al13 For the current study, data were used from the second wave of 1995 onwards because this was the first year that an identical disability question was asked from the participants in all countries. Countries for which mortality data (the Netherlands) and disability data (Luxembourg) were not available were omitted from the study. Germany and the UK were replaced by data from national surveys (German Socio-Economic Panel (SOEP) and British Household Panel Survey (BHPS), respectively). Austria and Finland joined the ECHP at wave 2 and 3, respectively. Table 1 presents information on cumulative retention percentages within the countriesdthat is, the cumulative percentages of individuals retained until the fourth wave of the panel. For most of the countries this was 1997; however, this year represented the third wave in Austria and the second wave in Finland. Individuals who were out of scope of the survey by 1997 (eg, because they died, became institutionalised or had moved outside the EU) were excluded from the calculation of these rates. There were large differences between countries, with relatively high retention percentages in Italy (83%) and Portugal (87%), whereas the samples in Ireland, Denmark and Spain suffered from high attrition.

Indicators The level of completed education at the first wave was used as a measure of SES. Individuals were divided into three groups according to their level of educational attainment based on the International Standard Classification of Education14: (1) lower secondary education or lower, (2) upper secondary education and (3) tertiary education, which is constituted by vocational and university education. All individuals were asked if they were hampered in their daily activities by any physical or mental health problem, illness or disability. The possible answers made a distinction between two severity degrees: ‘Yes, to some extent’ or ‘Yes, severely’.15 For this study, these persons were considered as a single ‘disabled’ category. Persons who were lost to the ECHP because of moving to an institution were also considered as ‘disabled’ for the remainder of the study period.

Table 1 shows the distribution of the elderly population by level of education of men and women. The countries with the most skewed distribution are the southern countries.

Data analysis In our multi-state Markov (MSM) models,16 17 three health states were defined: non-disabled, disabled and dead. Between the health states four transitions could occur: incidence (from non-disabled to disabled), recovery (from disabled to nondisabled) and state-specific mortality. For each country, transition rates were estimated on the pooled dataset controlling for age, gender, education level and the country of interest. In other words, a model for each country was specified whereby the country of interest was compared to the other nine countries. The advantage of preparing estimates using the data of all 10 countries together was the large number of observations; this ensured a high level of accuracy of the estimates of age profiles. We tested whether including two-way or three-way interaction terms between gender, education level and country of interest would result in a better model fit. Akaike Information Criteria (AIC) values indicated that the best model fit was achieved by including all two-way interactions but none of the three-way interactions. Because in most of the countries mortality cases were underregistered in the ECHP, the mortality transition rates were adjusted to the level of the given countries in four steps. First, using national data, mortality rates were calculated for the pooled period 1995e2001 for each age, gender and country. Second, they were transformed into mortality rates of nondisabled and disabled assuming that (1) the age and sex-specific mortality rates in the overall (ie, mixed non-disabled-disabled) population are the weighted average of mortality rates of nondisabled and disabled populations, with the proportion of nondisabled and disabled, respectively, as weights; and (2) that the ratio between the mortality rate of disabled and non-disabled people is equal to the HR as estimated with the ECHP data.18 A more formal explanation of the decomposition is shown in the appendix. Third, rescaling factors were calculated for both types of mortality rates by age and gender. Rescaling factors specified how much age-specific and gender-specific mortality rates had to be scaled to make them consistent with the national data. Fourth, all estimated mortality rates were multiplied by the corresponding rescaling factors assuming that under-representation of mortality was the same for all educational levels. To provide estimates of LE and DFLE according to gender, educational level and country, multi-state life tables were used.

Table 1 Total number of cases in the first wave, cumulative retention rates until fourth wave and proportion of education level in the study population, all countries Study population Men

Finland Denmark Ireland Austria Belgium Greece Italy France Spain Portugal

Women

Total number of cases at the first wave

Retention percentage until the fourth wave (%)

High level education (%)

Middle level education (%)

High level education (%)

Middle level education (%)

11 184 7537 14 170 9450 8976 15 872 21 520 18 643 22 578 14 285

92 67 64 82 78 77 83 75 71 87

28.6 35.2 21.0 62.2 28.3 13.6 17.6 29.2 7.2 2.8

21.9 31.7 10.5 5.5 26.6 10.2 6.3 15.6 9.7 3.3

25.9 26.2 23.2 38.6 24.6 8.6 12.2 21.4 4.2 1.8

22.9 24.3 7.9 2.8 18.4 4.1 2.6 10.7 4.4 2.9


973

Ageing The empirical input for these multi-state life tables were the transition rates described above after converting them into probabilities.19 Once a life table was set up for a country, probabilistic sensitivity analysis20e22 was performed to estimate CIs around the life expectancies. Random regression coefficients were drawn from each regression model for 1000 times assuming multivariate normal distribution. After each draw, the corresponding transition rates and life expectancies were calculated. The 25th and 975th of the latter ordered values indicated the boundaries of the CIs. All MSM analyses were performed in R,23 whereas life table calculations were carried out in Excel.

RESULTS The HRs of transitions by educational level for the 10 countries are given in table 2. Generally, the risk of disability onset and mortality from a non-disabled state was higher for the middle or low educated groups than for the high educated groups. For recovery rates, the higher the educational level the higher the rate to recover from disability. There were no differences between HRs of mortality from a disabled state. Educational differences in HRs were similar for both genders with some variation between the individual countries. The estimates of partial DFLE per country, gender and educational levels are shown in figure 1. Educational inequalities existed in partial DFLE in all countries. On average, people with a higher educational level can expect to live the most years without disability between the ages of 50 and 65 years. Partial DFLE was 10.1 years for high educated men, 8.9 years for middle educated men and 8.0 years for low educated men; the corresponding values for women were 10.8, 10.0 and 8.9 years. When comparing high to lower educational levels, these partial DFLEs translated into 2.1 and 0.9 years difference for men, and 1.9 and 1.1 years for women. Larger inequalities were found in Portugal and France and smaller inequalities in Belgium. Table 2

Table 3 presents LE estimates at age 65 years. A common pattern emerged in all countries: the higher the educational level the greater the LE at age 65 years. On average, the difference in men’s LE between high educated and lower educated groups was around 3 yearsdthat is, about double the difference between the middle and low educated groups, which is 1.3. Among women, differences in LEs were smaller than among men. On average, high and middle educated women can expect to live 1.9 and 1.3 years longer, respectively, than low educated women. Relatively larger differences were found in Austria and Portugal compared to smaller differences in Denmark and Finland. Table 4 presents estimates on the DFLE at age 65 years. The higher the educational level the longer people remain healthy at older age. For both men and women, differences between the educational attainment groups were larger for DFLE than for LE. Among men, the average DFLE difference between high and low educated groups was 4.6 years compared to 2.1 between middle and low educated men; for women, the corresponding values are 4.4 and 2.7, respectively. Larger differences were seen in Spain and Portugal, and smaller differences in Finland, Denmark and Belgium.

DISCUSSION This study explored educational differences in (disability free) LE before and after formal retirement age across 10 Western European countries. Populations with a higher level of education can expect to live more years free of disability before retirement suggesting fewer problems in reaching pension age in good health. People with a higher educational level can also expect to live longer after retirement implying that they represent a greater liability for pension funds. For LE, larger inequalities were found among men than among women whereas differences in DFLE were similar for both men and women. Similar patterns emerged in all countries, although the inequalities tended to be larger in southern countries.

HRs of disability related transitions in relation to educational level, by country Mortality of non-disabled

Mortality of disabled

Country

Incidence Middle

Low

Recovery Middle

Low

Middle

Low

Middle

Low

Men Finland Denmark Ireland Austria Belgium Greece Italy France Spain Portugal

1.15 1.19 1.48* 1.91* 0.97 1.39* 1.27 1.49* 1.72* 1.17

1.29* 1.37* 1.68* 1.99* 1.19 1.84* 1.86* 1.56* 2.86* 2.25*

0.74* 0.93 0.85 1.07 0.76* 0.78 1.03 0.74* 1.06 1.05

0.69* 0.76* 0.70* 0.88 0.78* 0.75* 0.81 0.55* 0.94 0.76

1.2 1.36 0.99 1.47 1.36 1.13 1.61 1.14 1.09 1.54

1.37 2.25 1.61 1.41 1.61 1.81 1.32 0.96 1.35 1.33

1.01 1.02 1.10 0.83 0.92 1.09 1.44 0.85 0.90 1.05

1.00 0.81 1.11 1.35 1.16 0.99 0.85 1.14 0.92 1.16

Women Finland Denmark Ireland Austria Belgium Greece Italy France Spain Portugal

0.97 1.00 1.25 1.60* 0.81 1.16 1.08 1.26* 1.44* 0.99

1.28* 1.33* 1.59* 1.95* 1.13 1.79* 1.83* 1.53* 2.75* 2.17*

0.71* 0.88 0.80 1.03 0.73* 0.75 0.99 0.71* 1.02 1.00

0.72* 0.79* 0.74* 0.91 0.83* 0.78* 0.85 0.58* 0.99 0.80

1.11 1.25 0.91 1.34 1.25 1.05 1.46 1.09 1.01 1.44

1.20 2.00 1.41 1.22 1.40 1.58 1.20 0.86 1.20 1.17

0.94 0.98 1.02 0.77 0.85 1.02 1.36 0.79 0.86 0.95

0.92 0.78 1.04 1.27 1.08 0.93 0.79 1.09 0.82 1.16

Derived from a multi-state Markov regression model that included age, sex, age 3 sex, middle level education, low level education, middle level education 3 sex, low level education 3 sex, country, country 3 middle level education, country 3 low level education, country 3 sex. *Significant at p¼0.05 level.

974


Ageing Figure 1 Partial disability free life expectancies (DFLEs) between the age of 50 and 65 years by educational level, for men and women separately.

Data issues Some limitations regarding the use of data need addressing. First, disability data were self-reported, which can result in either under reporting or over reporting of disability. If the reporting of disability differs by educational level then the differences in DFLE estimates will also be biased. Although we cannot exclude that reporting of disability differed by educational level, it is unlikely that differential reporting patterns would have strongly biased our results. To support this conclusion, an earlier study showed that educational inequalities in the prevalence of disability remained about the same when self-reported measures of disability were replaced by performance-based measures.24 Non-response and attrition might have been a problem in the present study if they had been related to disability and SES. A few studies have explored the association between attrition and disability status in the ECHP. For example, analyses on attrition in the ECHP showed a positive relationship with worsening health in all countries25 and only a weak relationship with educational level.26 Also, we assessed the likelihood of attrition in the current study population in relation to characteristics at the last wave in which the respondent participated.27 We found that the risk of loss to follow-up for reasons other than death or institutionalisation was hardly related to disability status or educational level. This implies that differential retention could J Epidemiol Community Health 2011;65:972e979. doi:10.1136/jech.2010.111492

not have strongly biased our estimates of relative educational differences in LE and DFLE. The distribution of educational level across the countries showed rather skewed patterns with large proportions in the lower educated groups. The sample size of higher educated persons was particularly small in Austria, Greece, Italy, Spain and Portugal. Because of the relatively small sample sizes, the CIs are wider for the higher educated groups. As a result, the CIs around LE and DFLE estimates for the groups with middle and high educational status often overlap. The same problem of random fluctuations might explain unexpectedly high or low values of LE or DFLE in some specific population as in the case of Italy. A main advantage of the ECHP was the availability of a large number of observations based on use of an identical survey design and survey questionnaire for all participating countries. Although the data collection was carried out by the National Institutes of Statistics separately and, hence, national versions of the ECHP are not perfectly comparable, these common questions ensured a much higher degree of comparability between countries than would have been possible using national data sources. The key variables used in this paper, on educational level and health status, are comparable across all ECHP countries albeit comparability of educational information could only be 975

Ageing Table 3

Total life expectancy (LE) at age 65 years by educational level, for men and women separately LE (95% CI) by educational level High

Middle

Low


15.7 15.8 17.8 17.5 16.2 19.2 19.0 19.4 19.4 18.7

Average

17.9 (17.2 to 18.5)

16.3 (15.9 to 16.6)

14.9 (14.8 to 15.1)

2.9 (2.5 to 3.4)/1.3 (1.2 to 1.5)


18.9 17.9 21.0 21.4 20.8 20.6 22.1 23.8 22.3 21.4

18.6 17.4 20.1 21.1 20.7 19.4 19.6 23.9 22.0 20.6

18.3 17.2 18.3 18.4 19.1 18.1 20.8 21.7 20.4 18.5

0.6/0.3 0.8/0.3 2.7/1.8 3.0/2.7 1.7/1.6 2.5/1.3 1.3/1.2 2.1/2.2 1.9/1.6 3.0/2.2

Average

21.0 (20.4 to 21.6)

(15.3 (15.4 (16.9 (16.6 (15.7 (18.5 (18.2 (19.0 (18.8 (17.8

(18.6 (17.5 (20.2 (20.7 (20.4 (19.9 (21.3 (23.4 (21.6 (20.7

to to to to to to to to to to


16.1) 16.2) 18.6) 18.3) 16.8) 19.8) 19.8) 19.8) 19.9) 19.5)

19.2) 18.4) 21.7) 22.1) 21.2) 21.2) 22.8) 24.2) 22.9) 22.2)

14.6 14.5 16.0 16.3 15.2 17.2 15.3 18.5 18.4 16.7

(14.3 (14.2 (15.5 (16.2 (14.9 (16.7 (15.0 (18.3 (17.9 (16.0

(18.4 (17.1 (19.7 (21.0 (20.4 (19.0 (19.2 (23.8 (21.5 (20.0



14.8) 14.7) 16.5) 16.5) 15.5) 17.6) 15.7) 18.7) 18.9) 17.3)

18.9) 17.8) 20.7) 21.2) 21.0) 19.8) 20.0) 24.1) 22.5) 21.2)

20.4 (20.0 to 20.7)

achieved at the level of three broad groups. Even so, international comparability may still be far from optimal due to crossnational variations in factors such as people’s perception of health problems and their propensity to report perceived health problems. Therefore, caution is needed when interpreting any differences in results between these European countries. Therefore, we recommend focusing on the patterns common to all countries rather than on the cross-national variations.

Previous studies Estimates of DFLE according to SES have been reported for several countries5e11 28e41; however, for Greece, Ireland and Portugal we believe that the current study is the first to report estimates of socioeconomic inequalities in DFLE. Data from other countries are not directly comparable with our data because of the different ages at which life expectancies were calculated or because different measures and classifications of disability and SES were used. A few estimates for SES differences in LE and DFLE at higher ages have been reported earlier. In the Netherlands, at age $65 years, for total LE and healthy LE a difference of 3.1 and 3.4 years, respectively, was found in men when comparing primary educated and higher educated groups.30 In England and Wales, in the age group $65 years, for estimated LE free of mobility limitations a difference of 2.7 and 2.5 years was found between high and low educated men and women, respectively; corresponding differences in LE were 1.1 and 1.9 years.5 In the USA, differences in LE between primary and higher educated were 2.5 and 3.3 years, respectively, whereas differences in active LE were 2.4 and 2.8 years for white men and women, respectively.37 For Italy, differences in DFLE were also found with 976

Highelow/middle-low

14.2 13.8 14.1 13.6 13.4 15.8 16.7 16.4 16.4 14.8

(14.0 (13.7 (13.9 (13.4 (13.2 (15.6 (16.5 (16.2 (16.3 (14.7

(18.2 (17.1 (18.1 (18.3 (18.9 (18.0 (20.7 (21.6 (20.3 (18.3



14.4) 13.8) 14.3) 13.9) 13.7) 15.9) 16.8) 16.6) 16.6) 15.0)

18.5) 17.2) 18.5) 18.6) 19.3) 18.2) 21.0) 21.9) 20.5) 18.6)

19.1 (18.9 to 19.2)

1.5/0.3 2.1/0.7 3.7/1.9 3.8/2.7 2.8/1.8 3.4/1.4 2.3/1.3 3.0/2.1 2.9/2.0 3.8/1.9

1.9 (1.5 to 2.4)/1.3 (1.1 to 1.5)

a 3-year difference between the higher and lower educated for both men and women at age 65 years.8 The present study shows similar basic patterns of DFLE and LE by SES. A common finding is that inequalities in HE were larger than inequalities in LE. Although comparison between studies is difficult, we found somewhat larger differences especially in terms of DFLE. This might be due to the milder measure for disability that was used in our study. At a pan-European level, previous studiesdimplicitly or explicitlydexamined the feasibility of increasing labour force participation around retirement eligibility age using health expectancy measures based on the ECHP data,42 43 the Statistics of Income and Living Conditions (SILC) data44 or the Survey of Health, Ageing and Retirement in Europe (SHARE) data.45 These studies assessed the average health expectancy and ‘unused capacity’ in several European countries and pointed at substantial differences therein. Our study extends this work for a selection of European countries included in the ECHP by focusing on variations by socioeconomic status. It would be particularly interesting to further extend our study to the Central-Eastern European (CEE) countries included in the SHARE or SILC surveys, even though SILC data do not contain information on mortality and SHARE data have only one wave with CEE countries. In previous studies, large inequalities were observed independent of the applied socioeconomic indicators (educational, occupational class, income or wealth measures). The inequalities that we observed in relation to educational level may reflect the operation of different causal mechanisms, including effects of socioeconomic position in later life. Blane46 proposed five explanations for the strong and persistent association between education and health in later life: (1) influence of childhood J Epidemiol Community Health 2011;65:972e979. doi:10.1136/jech.2010.111492

Ageing Table 4 Life expectancy without disability (DFLE) at age 65 years by educational level, for men and women separately DFLE (95% CI) by educational level High

Middle

Low

5.9 8.2 11.3 9.1 9.6 12.1 12.5 9.4 13.1 12.2

5.2 6.5 9.1 7.2 8.1 9.9 11.4 7.3 9.8 7.7


7 .8 9.7 14.2 11.8 11.0 15.4 15.8 13.0 15.2 14.0

Average

12.8 (11.5 to 14.1)

10.3 (9.5 to 11.2)


8.6 9.4 16.2 13.9 13.5 15.9 17.5 14.8 16.5 14.3

7.2 8.7 13.9 11.6 12.8 13.5 15.5 11.6 14.8 13.7

Average

14.0 (12.5 to 15.5)

(7.1 to 8.7) (8.7 to 10.7) (12.5 to 15.7) (9.8 to 13.7) (10.1 to 12.2) (14.1 to 16.5) (14.4 to 17.1) (12.0 to 14.0) (14.1 to 16.4) (12.1 to 15.7)




12.3 (11.3 to 13.4)

living circumstances on adult health, (2) effects of occupation and income achieved in adult life, (3) receptivity and adaptability to health education, (4) effects of ill-health during childhood on education and (5) influence of other background variablesdfor example self-efficacy time preference, and so forth. Although the evidence regarding cross-national variations is rather weak, it is interesting to find that inequalities in DFLE were generally larger in southern than in northern countries. Our results are generally in line with the hypothesis that in Nordic countries (where egalitarian welfare regimes have been implemented) inequalities in health may be smaller. Protective welfare systems could lead to smaller inequalities through their effect on income and wealth, on working conditions and on socio-psychological resources available to different socioeconomic groups. Nevertheless, other studies provide no consistent evidence that socioeconomic inequalities in mortality or selfreported health are smaller in Nordic countries.47

Highelow/middle-low (4.8 to 5.7) (6.0 to 7.1) (8.6 to 9.6) (6.7 to 7.7) (7.6 to 8.7) (9.6 to 10.3) (11.1 to 11.8) (6.9 to 7.8) (9.4 to 10.1) (7.4 to 8.1)

8.2 (7.8 to 8.7)

6.0 6.9 11.4 9.2 11.1 10.9 13.3 8.7 10.8 8.0


9.6 (9.2 to 10.1)

2.6/0.7 3.2/1.6 5.1/2.2 4.7/1.9 2.9/1.5 5.4/2.2 4.3/1.1 5.6/2.0 5.5/3.4 6.2/4.4 4.6 (3.7 to 5.4)/2.1 (1.7 to 2.5)

2.5/1.2 2.6/1.8 4.7/2.5 4.7/2.4 2.3/1.6 5.0/2.6 4.2/2.2 6.1/2.9 5.6/4.0 6.3/5.7 4.4 (3.3 to 5.4)/2.7 (2.1 to 3.3)

resources. On the other side, those with a higher educational level live more years in good health before reaching pension age. Although being disabled does not necessarily mean being unable to work, and being non-disabled does not necessarily mean being able to work, good health is associated with increased likelihood of participation. Our results indicate that in the case of lower socioeconomic groups increasing the retirement ages will be more difficult to achieve (given higher disability prevalence) or

What is already known on this subject < Life expectancy has been increasing steadily for many

decades in Western-European countries. < Life expectancy and healthy life expectancy are both strongly

related to socioeconomic status.

Conclusions Systematic reforms aimed at increasing pension(able) age have been proposed or implemented to take into account the trend of rising life expectancy and similar rises in health expectancy. However, such restructuring rarely acknowledges the differences in life and health expectancies between socioeconomic groups. For Europe at large, this study has shown that such inequalities are substantial in every country investigated. Educational inequalities favouring the higher educated exist on both sides of the official retirement age. On the one side, retired people with a higher educational level live healthier and longer lives and represent a subpopulation making greater demands on pension J Epidemiol Community Health 2011;65:972e979. doi:10.1136/jech.2010.111492

What this study adds < Western-European populations with a higher educational level

expect to live more years free of disability before retirement. < People with a higher educational level can also expect to live

longer after retirement. < If the official retirement age is raised, people at the bottom of

the socioeconomic ladder could be disproportionately affected.

977

Ageing to justify (given shorter life expectancies). Social policies should be oriented towards promoting the employment of seniors with higher SES because this is the group of people where the ‘unused capacity’ is largest.45 As a consequence, more flexible pension schemes could be considereddfor instance. taking into account the number of years worked over life course or allowing for part time pensioning. Acknowledgements MicMac has been developed by a team of researchers from the Netherlands Interdisciplinary Demographic Institute (NIDI), The Hague; the Vienna Institute of Demography (VID), the Institut National d’Etudes Demographiques (INED), Paris; the Bocconi University, Milan; the Erasmus Medical Centre, Rotterdam; the Max Planck Institute for Demographic Research, Rostock; the International Institute for Applied System Analysis (IIASA), Luxemburg, and the University of Rostock. Funding This work was supported by MicMac: an international research project funded by the European Commission in the context of the Sixth Framework Programme (grant number: SP23-CT-2005-006637). The funding organisation did not participate in the design and conduct of the study, collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript. This study was also part of the project ‘Living longer in good health’, which was financially supported by Netspar (grant number: 2007.3900.027). Competing interests None. Contributors AEK and IMM planned the study, developed and refined the methodological approach. IMM performed all statistical analysis and wrote the paper. All authors discussed and substantiated the interpretation of the results. AEK helped to revise the paper together with WJN and JPM. Provenance and peer review Not commissioned; externally peer reviewed.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

978

Queisser M, Whitehouse E. Pensions at a Glance: Public Policies Across OECD Countries. Paris: OECD, 2007:208. Whitehouse E. Life expectancy risk: Who bears the burden? OECD Social, Employment and Migration Working papers. Paris: OECD, 2007:46. Munnell AH, Soto M, Golub-Sass A. Will people be healthy enough to work longer? Boston: Center for Retirement Research at Boston College Hovey House, 2008:40. Lafortune G, Balestat G, Study D, et al. Trends in severe disability among elderly people: assessing the evidence in 12 OECD countries and the future implications. OECD Health Working Papers. Paris: OECD, 2007:80. Jagger C, Matthews R, Melzer D, et al. Educational differences in the dynamics of disability incidence, recovery and mortality: Findings from the MRC Cognitive Function and Ageing Study (MRC CFAS). Int J Epidemiol 2007;36:358e65. Matthews RJ, Jagger C, Hancock RM. Does socio-economic advantage lead to a longer, healthier old age? Soc Sci Med 2006;62:2489e99. Melzer D, McWilliams B, Brayne C, et al. Socioeconomic status and the expectation of disability in old age: estimates for England. J Epidemiol Community Health 2000;54:286e92. Minicuci N, Noale M. Influence of level of education on disability free life expectancy by sex: the ILSA study. Exp Gerontol 2005;40:997e1003. Nusselder WJ, Looman CW, Mackenbach JP, et al. The contribution of specific diseases to educational disparities in disability-free life expectancy. Am J Public Health 2005;95:2035e41. Sihvonen AP, Kunst AE, Lahelma E, et al. Socioeconomic inequalities in health expectancy in Finland and Norway in the late 1980s. Soc Sci Med 1998;47:303e15. Valkonen T, Sihvonen AP, Lahelma E. Health expectancy by level of education in Finland. Soc Sci Med 1997;44:801e8. Peracchi F. The European community household panel: a review. Empir Econ 2002;27:63e90. Huisman M, Kunst AE, Mackenbach JP. Educational inequalities in smoking among men and women aged 16 years and older in 11 European countries. Tob Control 2005;14:106e13. UNESCO. International Standard Classification of Education 1997. 1997. Eurostat. ECHP UDB - Description of variables. Data Dictionnary, Codebook and Differences between Countries and Waves. Luxemburg: Eurostat, 2003. Kalbfleisch JD, Lawless JF. The analysis of panel data under a Markov assumption. Journal of the American Statistical Association 1985;80:863e71.

17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

28. 29. 30. 31. 32. 33.

34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47.

Jackson C. Multi-state modelling with R: the msm package. Cambridge, UK: Medical Research Council Biostatistics Unit, 2007. Nusselder WJ, Looman CW, Marang-van de Mheen PJ, et al. Smoking and the compression of morbidity. J Epidemiol Community Health 2000;54:566e74. Nusselder WJ, Peeters A. Successful aging: measuring the years lived with functional loss. J Epidemiol Community Health 2006;60:448e55. Briggs A, Schulpter M, Claxton K. Decision Modelling for Health Economic Evaluation. Oxford: Oxford University Press, 2006. Drummond MF, O’Brian BJ, Stoddart GL, et al. Methods for the economic evaluation of health care programs. Oxford: Oxford University Press, 2003. Boshuizen HC, van Baal PH. Probabilistic sensitivity analysis: be a Bayesian. Value Health 2009;12:1210e14. Team RDC. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2008. Huisman M, Kunst A, Deeg D, et al. Educational inequalities in the prevalence and incidence of disability in Italy and the Netherlands were observed. J Clin Epidemiol 2005;58:1058e65. Eurostat. Attrition in the ECHP. Luxemburg: Eurostat, 2002. Eurostat. Sample attrition between wave 1 and 4 in the European Community Household Panel. Luxemburg: Eurostat 2002. Majer IM, Nusselder W, Kunst AE. Age profiles of mortality and disability-related transitions in European countries. Estimates according to smoking status, and overweight status, educational level and marital status. MicMac reports. Rotterdam: Erasmus MC, 2008:58. Wilkins R, Adams OB. Health Expectancy in Canada, Late 1970s: Demographic, Regional, and Social Dimensions. Am J Public Health 1983;73:1073e80. Boshuizen HC, Van De Water HPA, Perenboom RJM. Sociaal-economishe verschillen in de gezonde levensverwachting. Tijdschr Soc Gezondheidsz 1994;72:122e7. Van De Water HPA, Boshuizen HC, Perenboom RJM. Health expectancy in the Netherlands 1983e1990. Eur J Public Health 1996;6:21e8. Doblhammer G, Kytir J. Social inequalities in disability-free and healthy life expectancy in Austria. Wien Klin Wochenschr 1998;110:393e6. Bronnum-Hansen H, Andersen O, Kjoller M, et al. Social gradient in life expectancy and health expectancy in Denmark. Soz Praventivmed 2004;49:36e41. Bebbington AC. Regional and social variations in disability-free life expectancy in Great Britain. In: Robine JM, Mathers CD, Bone MR, et al, eds. Calculation of Health Expectancies: Harmonization, Consensus Achieved and Future Perspectives.London: John Libbey Eurotext,1993. Cambois E, Robine J-M, Hayward MD. Social inequalities in disability-free life expectancy in the French male population, 1980-1991. Demography 2001;38:513e24. Bossuyt N, Gadeyne S, Deboosere P, et al. Socio-economic inequalities in health expectancy in Belgium. Public Health 2004;118:3e10. Van Oyen H, Bossuyt N, Deboosere P, et al. Differential inequity in health expectancy by region in Belgium. Soz Praventivmed 2005;50:301e10. Guralnik JM, Land KC, Blazer D, et al. Educational status and active life expectancy among older blacks and whites. N Engl J Med 1993;329:110e16. Crimmins EM, Hayward MD, Saito Y. Differentials in active life expectancy in the older population of the United States. J Gerontol B Psychol Sci Soc Sci 1996;51: S111e20. Lievre A, Alley D, Crimmins EM. Educational differentials in life expectancy with cognitive impairment among the elderly in the United States. J Aging Health 2008;20:456e77. Martıńez-Sańchez E, Gutie´rrez-Fisac JL, Gispert R, et al. Educational differences in health expectancy in Madrid and Barcelona. Health Policy 2001;55:227e31. Davis P, Graham P, Pearce N. Health expectancy in New Zealand, 1981e1991: social variations and trends in a period of rapid social and economic change. J Epidemiol Community Health 1999;53:519e27. Lievre A, Jusot F, Barnay T, et al. Healthy working life expectancies at age 50 in Europe: a new indicator. J Nutr Health Aging 2007;11:508e14. Christensen K, Doblhammer G, Rau R, et al. Ageing populations: the challenges ahead. Lancet 2009;374:1196e208. Jagger C, Gillies C, Moscone F, et al. Inequalities in healthy life years in the 25 countries of the European Union in 2005: a cross-national meta-regression analysis. Lancet 2008;372:2124e31. Borsch-Supan A, Brugiavini A, Croda E. The role of institutions and health in European patterns of work and retirement. J Eur Soc Policy 2009;19:341e58. Blane D. Commentary: explanations of the difference in mortality risk between different educational groups. Int J Epidemiol 2003;32:355e6. Mackenbach JP, Stirbu I, Roskam AJ, et al. Socioeconomic inequalities in health in 22 European countries. N Engl J Med 2008;358:2468e81.


Ageing APPENDIX Mortality decomposition Raw total mortality rates for the pooled period 1995-2001 were calculated for each age and sex based on the total counts of mortality and exposure time during the period of interest, as published in Human Mortality Database (HMD). Age- and sexspecific prevalence of disability measures were calculated based on the number of disabled persons and the number of participants in the study population during the same period as published in the online database of European Health Monitoring Unit. Age- and sex-specific (but not country-specific) hazard ratios were estimated using MSM models based on the pooled study populations.

Combining total mortality rates, hazard ratios and prevalence of disability, mortality rates were calculated for non-disabled and disabled populations by the following formulas:(1) ðndÞ

mx

ðdÞ

¼

mx ¼ ðndÞ

ðdÞ

mx ðdÞ ðdÞ HRx 3px þ ð1px Þ ðndÞ mx 3HRx

(1)

ðdÞ

where mx , mx , HRx and px indicated the mortality rate of non-disabled, mortality rate of disabled, estimated hazard ratio and prevalence of disability at age x, respectively.

Advancing Postgraduates. Enhancing Healthcare.

ES READSENTIAL ING PLA FOR

EXAM B INEES

The Postgraduate Medical Journal is dedicated to advancing the understanding of postgraduate medical education and training.

• Acquire the necessary skills to deliver the highest possible standards of patient care • Develop suitable training programmes for your trainees • Maintain high standards after training ends Published on behalf of the fellowship for Postgraduate Medicine FOR MORE DETAILS OR TO SUBSCRIBE, VISIT THE WEBSITE TODAY

postgradmedj.com


979